THE BOOK--Playing The Percentages In Baseball

I can see how Brock would look better.  First, the SB success rate was quite a bit lower in his time: around 63% in his prime years, compared to about 69% for Raines.  So that makes his batteries look “better,” although it may just be that teams made worse decisions about sending runners.  And Brock (or his managers) may also have made worse decisions about when to go.  For example, 28% of Brock’s attempts were against LHPs, vs. 23% for Raines.  Whether that makes Brock a “better” basestealer, though, I think is debatable.

Guy/1, my point is similar to your second point.  Is there a certain level of talent in deciding when to run?  For example, maybe Brock is mostly concerned with running when his team could really use the base, though he’d also run on very bad batteries.  And maybe Raines would also consider the situation, but he had a better knowledge/eye for when a battery was below average.  Thus he would run more often against weaker batteries because he’d run on most below average batteries the way Brock would run on very bad batteries.  Some of that may even be the effect of a good 1B coach.  It’s a thought, and I’m not sure how feasible it would be to isolate SB opportunities by battery and game state to judge comparable choosing-when-to-run skills.

WOWY might break down to some extent for baserunners across different eras because the league CS% has dropped quite a bit over time due to strategical changes in decision-making.  There’s a good chance that teams were running more recklessly once upon a time and have deliberately picked out fewer low-success attempts as time has gone on.  For example, let’s say the batteries Brock is being compared to are being charged with throwing out any idiot who wants to run even if they have less than a 50%.  So say the pool of batteries Brock is running off of are facing runners running in the following situations (in terms of how likely they are to succeed):

40%, 45%, 50%, 55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%

and let’s say the average success rate is around 62%.  Let’s say these runners are running 36% of the time (I’m completely making that up).  They’d average about 22 steals and 14 caught stealings per 100 chances, which would set a baseline, using the +.3/-.6 weights, of about -1.5 r100.

Now let’s say that runners decide that they should not be running in 50% or worse situations, and now they only run in the following:

55%, 60%, 65%, 70%, 75%, 80%, 85%, 90%

And because they are not running in low success opportunities, they now run at a 67% rate but only run 30% of the time.  They’ll have a baseline right around zero r100.

Now assume that the pools of runners and the pools of batteries in each of these two groups is the exact same, except that in the second group, runners stopped trying to steal in low-success situations.  When you run your WOWY on catchers, it will tell you that this pool of baserunners is harder to throw out because they have a higher success rate, which it is.  That’s not a problem.  But when you run your WOWY on baserunners, it will tell you that the batteries got worse at throwing runners out, so it is easier to steal off of them.  That is not true.  The batteries stayed the same, but the offense deliberately removed the worst baserunners from the sample, so instead of picking up that Brock is running on harder catchers, what you are picking up is that he is being compared to a worse pool of baserunners.

I’m not sure how much that is affecting the Brock/Raines comparison once you look at real numbers, but I would bet that is part of the issue, possibly to the point that WOWY won’t work for this question (or at least not without tweaks).

Repeating what was said in 1 and 2 (which it seems I’ve already done), that’s another reason WOWY might break down for baserunners on stolen base attempts, depending on how you do it.  Base-stealers are free to choose what batteries they do and don’t steal off of, so if Brock chooses to steal off or harder batteries (like running off of more lefties), then that means he really is running off or harder batteries, but it doesn’t make him a more valuable base-stealer because another base-stealer could be in the same situation and simply choose not to run off of that battery.

If you are measuring the quality of batteries faced by only looking at when a baserunner attempts to steal, this will be an issue.  Runners who do choose to run in bad spots will be deflating their actual value, but your measurements might not pick up on that because they are adjusting the baseline down as well.  If, however, you are looking at every battery when a runner is on first (and second and third open) and asking how often will the average runner attempt to steal given this mix of batteries, and how many stolen bases and caught stealings will he end up with? then you might be able to avoid the issue.  If the average baserunner stays put against a Pettitte/I-Rod combo, and you decide to run, it doesn’t affect your baseline because that was being counted in your mix anyway, and when you get thrown out, it is not compared to the 60+ percent chance that you were going to get thrown out once you went so that you only take a small hit, it is compared to the 95% chance that another runner would have just stayed put and lost no value, so you take a big hit.  So it depends on how the mix of batteries is computed whether this will be an issue.  It sounds like Max may have done it looking at the whole mix of batteries in running situations, and not just at the mix of batteries you run off of, so that issue may have been avoided here.

This is a fantastic point guys.

Really, what we are measuring is: GIVEN that he decided to steal, how successful is he against other runners who ALSO decided to steal.

From that standpoint, it’s more than believable that Brock = Raines.

It’s just that Raines attempted when the breakeven point was higher than when Brock attempted.  The overall effectiveness obviously favored Raines in the end.

Just a fascinating point and one I should have been able to see.  Great job guys.

So, we may actually be in the position to split up a runner’s effectiveness between his straight basestealing and his choosing to running.  That is very cool.

For example, if we find that Raines = Brock, then I would be able to fill out my chart here:
http://www.insidethebook.com/ee/index.php/site/comments/chase_utley_sb_wizard/

So, we can look at high success / low attempt guys like Carlos Beltran and put him in context with other basestealers.

Very very cool…

There’s a great article/study to be written about the big gain in baserunning efficiency over the past 40 years, and especially the sharp increase over the past decade or so. 

First, how have teams achieved this?  More selective attempts by good basestealers (e.g. Beltran)?  Fewer attempts by weak runners?  Both?

Have teams also gotten smarter about WHEN they run (based on score, inning, count), or just achieved higher success rates across the board?

How much of this was just a reaction to increased HR levels, which lowers value of SBs, as opposed to a new understanding of the cost of CSs?  And if the latter, what role did saber analysis play in that (if any)?

"The overall effectiveness obviously favored Raines in the end.”

I imagine that’s true, but maybe not as much as you think.  In Brock’s era, I imagine gaining an extra base was a bit more valuable.  More important is the issue we discussed in the catcher thread:  Brock’s aggressive running also had a payoff on many plays when StL hitters put the ball in play.  I bet he went first-to-third a lot on singles, even singles not hit to RF, and avoided a lot of GDPs.

The win probability numbers (which takes into account the game situation) greatly favor Raines to an extent that he’s a shade behind Rickey Henderson in basestealing.